Acute respiratory failure requiring mechanical ventilation is an enormous clinical and financial burden on the health system, affecting over 750,000 patients in the United States each year. These patients experience substantial morbidity and mortality, particularly in small hospitals with low annual caseloads. To improve outcomes for these patients, many experts propose a regionalized approach to mechanical ventilation, analogous to the United States trauma system. Under regionalization, mechanically ventilated patients admitted to small, resource-poor hospitals would be systematically triaged to large regional referral hospitals. Regionalization has potential to save lives and increase access to high-quality critical care for patients requiring mechanical ventilation. Yet, implementation efforts face significant barriers, most notably the lack of empirical data demonstrating that centralizing care in regional hospitals of excellence actually improves outcomes. Although many studies demonstrate that mechanical ventilation outcomes are better at high-volume hospitals compared to low-volume hospitals, these studies fail to fully uncover the implications of regionalized care, which involves not only increasing case volume at some hospitals but also decreasing it at others. Prior to moving forward, clinicians and policy makers require empirical data on the implications of regionalized critical care in real-world settings. The overall goal of this project is to empirically evaluate the outcome benefit of regionalized critical care for patiets with acute respiratory failure. In Aim 1 we will define novel critical care referral regions as a geographic foundation for quantifying regional care delivery. In Aim 2 we will identify regional factors associated with greater centralization of care for acute respiratory failure. In Aim 3 we will determine the association between centralized critical care and outcomes for patients with acute respiratory failure. Completion of these Aims will both advance our knowledge of the relationship between regionalization and outcomes for acute respiratory failure, as well as provide a scientific foundation for future efforts to centralize care for other high-risk patients.The research plan will be augmented by intensive mentoring by experts in the field and didactic research training at the University of Pittsburgh and Carnegie Mellon University. Together, the research project, mentoring and coursework described herein will provide the primary investigator with essential career development in the areas of: (1) spatial epidemiology and health care geographical information systems analysis, (2) advanced hierarchical statistical modeling and (3) organizational science and health care market analysis. Ultimately, this work will set the stage for research evaluating centralized care of other time-sensitive conditions in a comprehensive R01-funded project and uniquely position the primary investigator as a future leader in the use of geographic modeling to improve outcomes for critically ill patients.